From owner-freebsd-hackers Sat Nov 14 15:48:11 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA27911 for freebsd-hackers-outgoing; Sat, 14 Nov 1998 15:48:11 -0800 (PST) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from smtp03.primenet.com (smtp03.primenet.com [206.165.6.133]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id PAA27897 for ; Sat, 14 Nov 1998 15:48:06 -0800 (PST) (envelope-from tlambert@usr02.primenet.com) Received: (from daemon@localhost) by smtp03.primenet.com (8.8.8/8.8.8) id QAA04768; Sat, 14 Nov 1998 16:47:34 -0700 (MST) Received: from usr02.primenet.com(206.165.6.202) via SMTP by smtp03.primenet.com, id smtpd004754; Sat Nov 14 16:47:30 1998 Received: (from tlambert@localhost) by usr02.primenet.com (8.8.5/8.8.5) id QAA28187; Sat, 14 Nov 1998 16:47:24 -0700 (MST) From: Terry Lambert Message-Id: <199811142347.QAA28187@usr02.primenet.com> Subject: Re: [Vinum] Stupid benchmark: newfsstone To: mike@smith.net.au (Mike Smith) Date: Sat, 14 Nov 1998 23:47:24 +0000 (GMT) Cc: ticso@cicely.de, mike@smith.net.au, peter.jeremy@auss2.alcatel.com.au, hackers@FreeBSD.ORG In-Reply-To: <199811132336.PAA01117@dingo.cdrom.com> from "Mike Smith" at Nov 13, 98 03:36:14 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > > > Most modern "server class" SCSI disks support it. It's not useful > > > unless you turn off tagged queueing, caching and most other drive > > > performance features. > > > > Where's the problem with these options on when using Spindle Sync? > > The whole point of spindle sync is to exactly lock all the drives > together to coordinate read/write activity. These features in > conjunction with sector sparing and quantum differences between disks > means that synchronising spindles is a complete waste of time, as the > disks won't be mimicking each other anyway. My personal take on this was to allow the choice of disk from which to read a stripe to be "late bound" to optimize the load distribution between spindles. This assumes each plex or stripe has at least one replicate on a different spindle. Give three read operations against three disks, this would allow you to choose your disks so as to schedule the reads from seperate disks for each of the reads on average 66% of the time. Even so, sync is much less useful inreverse sector ordered devices for which the geometry is not known, even for this use. Actually, you should drag Rod Grimes into any spindle-sync discussion; I know that he built a spindle-syned system, and had, at one time, put up patches for CCD to deal with it. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message